Steve Wing

Thank you Dr. Lurier. Thank you to the Star Foundation for the invitation to be here. I want to begin by introducing my colleague David Richardson and just to give you a little background, we have been involved in studies of workers at Department of Energy facilities. Ah, these studies most recently have been funded by the National Institute for Occupational Safety and Health. We've been studying workers from Oak Ridge, Savanna River, Los Alamos, and Hanford.

Today, we'd like to talk primarily about our studies at Oak Ridge. And it's fitting that Dr. Morgan is here because in the early 1940's he began the radiation protection programs that provided us with the information that today is used for long term studies of the health of these workers. IT's also fitting that Dr. Stewart is here. She worked beginning in the 1970's with Dr. Mancuso who in the 1960's began to assemble the kinds of records ah, that were made available to us in the early days of Health Physics, ah, and which are now the basis of--of the studies we'll discuss.

Ah, David will talk to you about some of the recent findings from our studies of Oakridge National Laboratory workers--ah, findings that have been produced over the last few years, and then I will discuss after David's finished some of the methodological issues involved in the design, analysis and interpretation of the worker data.

David Richardson

Thanks. It's a pleasure to be here at a meeting that includes so many people who are organizing to ensure their own safety, the health of their communities, and the health of future generations who will inherit the contamination that's been created by our military technologies. If I could have the first and second slide. Thanks. And a little bit of focus--not only for myself.

The research that I'm going to talk about concerns, ah, workers at Oakridge National Laboratory. The laboratory was originally a secret facility that was constructed in rural Tennessee as part of the Manhattan Project, which was the U.S. government's program during World War II to build an atomic bomb. Oakridge was the site of one of the world's early nuclear reactors, and since the 1940's has been involved in ah, research on energy related topics. As Steve mentioned, the facility provides a lot of valuable information for understanding the long term effects of exposures to low levels of, ah, ionizing radiation. One reason is that workers at Oakridge were monitored for external exposure to ionizing radiation by using primarily film badges that they clipped on to their clothing. So, over time these have been collected and provide a historical record of radiation exposures on the job.

The experience is also relevant for people who are interested in worker safety, ah, because employees at Oakridge tended to receive repeated low level exposures that were at levels that are typical of what many workers in the ah, nuclear industry today will be receiving. Overall, cancer death rates at Oakridge are less than cancer death rates in the U.S.. However, this is typical of what you find when you compare the general population that's going to include the chronically ill, the unemployed, the poor, the uninsured to a group of workers, most of whom are highly educated, well paid, employed by or recruited from top corporations and universities across the country. So, what you're seeing when you're looking at a low overall mortality or o--low overall cancer death rates is the fact of social inequalities--that there's differences in living conditions, access to medical care, employment, and so on that leads workers at Oakridge to have low cancer death rates.

Despite the low overall cancer mortality rates among workers at Oakridge National Laboratory, however, there have been more deaths due to leukemia than expected. The graph up here shows leukemia death rates among white male workers at Oakridge National Laboratory compared to leukemia death rates for white males in the U.S. as a whole. We singled out white males here since worke--since these are the workers that had the most complete vital status follow-up over time and these are the workers who receive the vast majority of radiation exposure on the job.

In the graph, the line indicates a comparison of Oakridge workers to the U.S. population, where values greater than 100% indicate that leukemia death rates among workers at Oakridge were higher than death rates in the U.S. population and values lower than 100% means th--there was a deficit of leukemia.

The pattern over time that you can see when you look at the graph is interesting to interpret in light of the trends in radiation exposures at the Oakridge facility. Average exposures at ONL increased progressively from 1940's through 1957 at which time they're at their peak and then they gradually begin to decline. And when we're looking at excess leukemia death rates, you see a similar trend with a short lag, where leukemia death rates were lower than expected in the early years of operation, increased over time and then in more recent years have begun again to decline.

OK. Taking that comparison of workers at Oakridge to the general population, just as a first observation, we went on then to look in more detail at the relationships between exposures to radiation on the job and mortality rates. I'm going to be talking primarily about 14,095 workers. These were people who were hired between 1943 and 1972 and who worked at least 30 days at the laboratory. We got information about these workers on their causes of death for all workers who had died as of 1990, and we got, again, information about the radiation exposures from their film badges that the workers were wearing. And then we used employment records to get information about each worker's date of birth, their gender, their race, their pay code, and whether they were also monitored for internal contamination from inhaling or ingesting radionucleids.

And we've used that--the employment i-information about those factors to adjust for differences that might occur between workers when we're focusing on interest in--differences in mortality rates with different levels of radiation exposure that they got on the job.

In looking at the relationship between radiation exposure and cancer mortality, one very consider--important consideration for us was the effects of aging. This is primarily because of the work that ah, Dr. Stewart's done, as well as the work of other people who have looked at radiation exposure among nuclear workers. It's also just a sensible question when you're looking at workers and exposures that begin on the job to ask the question, as you get older, are there changes in your sensitivity or your vulnerability to the effects of occupational exposures. As--as all of us get older, our body starts to slow down, our immune system starts to decline, the efficiency of healing and repair processes declines, and since those immune repair functions are believed to be an important part of mitigating the effects of exposure to ionizing radiation, then you can ask, well, given those changes, might the effects of radiation also change with age?

OK, so in the analyses that I'm going to talk about the way that we considered age at exposure was to look at radiation doses that the workers received on the job in the early years of their age and the later years of their age. For example, if you imagine a worker who started employment at Oakridge ah, at age 30 and continued employment until age 55, along the way the worker's going to receive exposures to radiation on the job ah, year after year potentially, low levels of exposure. So we separately counted up the radiation doses that a worker'd receive at their young ages, ah, for example, before age 45, and the radiation doses that they received in their later period of age. And then we looked to see if the effects of radiation doses received at younger and older ages differed--differed.

OK, with that in mind I'm going to show you three tables of results. In each table the results are described as the average increase or the average change in death rates with increasing radiation exposures.

I want to go over the first table rather slowly, just to get the hang of what we're looking at. As the text says, the numbers in the table describe the percentage change in cancer death rates per 10 [milliceverts?] or as we've been asked to say, one rem, which is 10 milliceverts. To put 10 milliceverts into context, ah, occupational exposure limits in th--for U.S. workers are 50 millicevrets per year. So as we're looking at the percentage change in a death rate per 10 millicevret, remember that a worker can receive five times that amount each year on the job under current regulations.

The title also says that the findings are under a 10 year lag assumption. What that means is that we've included all the radiation doses that a worker received up until ten years prior to their death. The reason that we're not looking at the la--the exposures that were received in the last years of their life would be that there's a common assumption is that there is going to be several years between any relevant exposure to radiation and a subsequent fatal cancer.

OK. In the first column of the table, we're looking at the changing cancer death rates per 10 millicevrets dose that's received at all ages. So if this is all the radiation dose, cumulative radiation dose under a ten year lag for a worker, on average you see that cancer death rates increased 1.35% per 10 millicevrets or 1 rem of radiation dose. This indicates that there's increasing cancer death rates with relatively small levels of radiation exposure that the workers are getting on the job.

What I want to focus on, though, are the next two columns. So, in these columns, we're breaking down the relationship between radiation exposure on the job and cancer death rates by looking at--separately at doses that were received in the younger periods of age and in the older periods of age. For doses that were received before age 45, there was little association with cancer death rates. This is indicated by a small estimated percent change relative, especially to the size of the standard error. In contrast, looking at the doses that are received after age 45, there's a substantial increase--4.9 percent increase in cancer mortality rates per 10 millicevrets dose.

Now, I'd mention here--we've taken age 45 to illustrate this. There's nothing particularly unique about age 45 as being some magic number where people suddenly become extremely vulnerable to radiation. Rather we c--we divided radiation doses looking for example at age 40, 45, 50, and you see the same overall pattern. At older ages there's a larger association between radiation exposure and cancer mortality.

I'm just going to very briefly talk about the numbers along the bottom of the table. These are statistical values that indicate how well the models that we're using descri--fit the observed data, where larger values indicated a better fit to the data. So these values indicate that the association between radiation and radiation received at all ages and cancer mortality was positive, but th--this was primarily due to the a--association when you're looking at the doses that are received at the older periods of age.

OK. In the next table, we're looking at radiation doses that are received after age 45, and changes in death rates under different lag assumptions. So as longer lag assumptions are considered, I'll remind you again, we're considering longer intervals between radiation exposure and any subsequent cancer. There's three rows in the table and they're looking at the association of radiation with all causes of death, all cancers, and all causes except cancer. Now when we went into this, we expected that if low level exposure to radiation was associated with changes in death rates among the workers at Oakridge, it was going to be primarily due, or most strongly due to associations between radiation and cancer--that is, when you're looking at all causes of death except cancer, you're including a--a big array of things that would include injuries, homicide, cardiovascular disease, infectious disease, and which if there was any association with radiation which you might not expect, it would be minimal at best.

So, in the first row you can see that radiation doses received after age 45 were associated with an increase in all causes of death under each of the three lag assumptions we're looking at. In the next row we're breaking this down. We're saying OK, let's break all causes down into all cancers and all causes except cancer. What you see is that radiation doses received after age 45 were strongly associated with cancer under each of the lag assumptions and there was much less evidence of an association between deaths due to causes other than cancer and radiation.

And as I promised you three tables--and this is the last table I want to talk about. In this table we've taken the broad category of all cancer deaths and we've broken it into two groups: lung cancers and cancers other than lung. One reason that you would consider lung cancer separately from other cancers is that there's a strong association between cigarette smoking and lung cancer, and so cigarette smoking would be v--expected to be very strongly associated with lung cancer and much less associated with other cancers. Then similar to the previous tables that I've shown, these analyses can be thought about as a way of looking for evidence of an alternative explanation for the relationship between cancer deaths and a radiation dose received after age 45. When we take lung cancer out of the associations--um, out of the group of all cancers--we expect that it would be much less plausible that cigarettes would explain any association between radiation and cancer. So we found that radiation was strongly associated with lung cancer and it was strongly associated with cancers other than lung, under 5, 10, and 20 year lags.

So, among workers at Oakridge National Laboratory we've seen that cancer death rates increase among workers who've received relatively small external exposures to ionizing radiation at levels that are well below current occupational standards. The association is primarily due to cancer causes of death rather than non-cancer causes or death, and it was similar in magnitude for lung cancer and cancers other than lung. And of particular interest, what we've seen is that the effects of exposure were largest for the workers who were getting exposed at older ages on the job.

This morning, Dr. Stewart suggested that the vulnerability to ionizing radiation is at its peak when we're at our very youngest--the foetus in the mother's womb, infants who are most sensitive of the damage caused by ionizing radiation, and then that s--sensitivity, she suggested declines in the middle age of life, in--in early adulthood and then begins to increase again with older age, at a period when our body's immune function or repair pro--processes are declining. This description should be familiar for most people who are thinking about injuries that you get, whether it's radiation or other types of injuries.

There's been a history in this field of authorities getting up and stating how much is understood about the effects of exposure to ionizing radiation and understating when they're talking to workers and the public the potential hazards of these exposures. I'd suggest that from this study, we can take these findings at best as cautionary observations. We're looking at the most o--extreme effects of a worker's exposure on the job--that it kills them--that is, we're not looking at non-fatal effects of radiation, um, the possibility of increasing cancers that are going to cause immense suffering, but aren't going to kill the worker, and we're not looking at ah, for example rep--human reproductive [p...?] effects. So this is one study about a very extreme consequence of getting exposed to radiation. What these findings do point to is a recognition that in any population there are likely to be substantial differences in sensitivity to the effects of radiation. We looked at aging as one variation between people and there's ah, very likely that there are genetic traits, pre-existing diseases, environmental exposures which all--might also cause substantial differences between groups of people.

Given the seriousness of the dangers that arise from nuclear technologies, I suggest that it's important to acknowledge that issues about the range of effects, the magnitudes of exposures and differences in vulnerability in a community are poorly understood. It's therefore appropriate for workers and the public who are faced with these hazards to be skeptical of anyone who gets up and is dismissive of their concerns. And it's inspiring to see people who are actively organizing to make their concerns heard. [applause]

Steve Wing

I want to briefly discuss some of the ah, interpretation that have widely been made of worker studies. I want to do this very quickly here. Um, by beginning with a few quotations from the [literature?]. And a--by these I want to--to draw attention to the way the study that Dr. Stewart described this morning--the Life Span Study of A-bomb survivors has been used as a context in which to interpret information presented about workers. Ah, if we can have the slides on please? I just press the button and--there we go.

Ah, so in the recent study which was cited earlier today by [Hal?] Morgenstern, ah, [...?...] state that a primary objective of studies of risk--cancer risk among nuclear industry workers is the assessment of the adequacy of the existing protection standards based on risk estimates based on risk estimates derived from analyses of the mortality of atomic bomb survivors. Similarly, ah, Kendall and colleagues studying British workers note that one of the objectives of studies of these workers is to obtain direct estimates of risk from exposures to low doses of radiation at low dose rates for comparison with the risk factors derived by the International Commission on Radiological Protection, mainly from high dose and high dose rate exposures of the Japanese atomic bomb survivors.

And finally, ah, from the United States, Gilbert et al. note that the analyses expressing risks and confidence limits this re--is referring to studies of Han--Hanford, Rocky Flats and Oakridge workers, ah, as multiples of the biological effects of ionizing radiation committee predictions ah, and they are provided in a table, which allowed for the--certain variations, provide stronger confirmation that risks have not been underestimated by high dose extrapolation.

The point I want to make is that the worker data have been interpreted in the context of the A-bomb survivor data. Let's briefly consider some of the design aspects of these two groups of studies. And remembering here the point David made that we're now talking about external radiation, not the internal emitters we heard about just a few minutes ago. And we're talking about cancer and not other outcomes.

Ah, first design issue I'd like to call your attention to is the sample--the life span study is a reconstruction of survivors in two cities, of all ages--

End of Side One

Day 1 Tape 2 Side 2

[two minutes from Side 1]

Ah, first design issue I'd like to call your attention to is the sample--the life span study is a reconstruction of survivors in two cities, of all ages whereas the worker studies are based on rosters, lists of workers who were employed and who were paid by these facilities. Ah, it's adults only, but rather than a, ah, reconstruction many years after the fact, five years after the fact ah, we have--we begin with a list which in the defined sample is an important aspect of epidemiological studies. The exposure ah, of A-bimb--A-bomb survivors primarily acute gamma and neutron doses from low levels to lethal levels, ah, the workers are exposed to the ac--to chronic low doses primarily gamma radiation, and this is more the situation we're interested in for purposes of radiation protection.

The exposure measurement is quite different in these two situations. Ah, for the A-bomb survivors the exposures are estimated based on physical models of what kinds of radiation were emitted by these two bombs, and survey responses of survivors many years after ah, a catastrophic experience. So, here we're dealing with modeled predictions, including in one case a bomb which ah--another one like it was never detonated, ah, and--and survey responses, whereas for the worker studies such as the ones David talked about ah, in Oakridge and ah, th--Hal Morgenstern discussed earlier--these are studies of people who wore [dosimeters?] which they put on when they entered the facility and took off when they left--not problem free, but individual measurement.

Selection factors are--ah, have been discussed in the case of the life span study--the A-bomb survivors by Dr. Stewart, ah, serious issues with respect to ah, selective survival of people at different ages and in different dose groups. Ah, worker studies also suffer from problems with ah, selection--we've heard about the healthy worker effect, in terms of SMRs--Standardized Mortality Ratios lower than the general population. So these are healthier people. They may be less sustep--susceptible than many other population groups to a variety of exposures including radiation. There are also important internal selection effects ah, that may influence dose response findings. Workers have to remain healthy and often they have to pass physical examinations and other selection processes in order to be exposed and to continue to be exposed. So we have a situation where we may ah, see that workers with higher exposures are a select group, healthier than workers with low exposures, and this may expect our dose response findings as well.

But the A-bomb survivors continue to d--to dominate the literature, um, as noted for example by [sounds like, Laurel Sieber...?...] in a paper, in analysis the models were chosen, used in this study, of paternal effects in the Hanford workers because of the models used for the A-bomb survivors.

Um, interpretation has been primarily based--and I would argue, inordinately influenced by statistical criteria having to do with precision and sample size. Ah, th--this quote from Kendall, and I'm going to--I know I have very little time, so I'm going to go rapidly here--ah, from the British studies, ah, indicates that the interpretation is being made in terms of statistical confidence limits, ah, and that samples--the very large sample size of the Japanese population, of course, ah, yields smaller ah, confidence limits. Ah, Gilbert, in the United States again, ah, sees results which differ in her analyses of worker data from the A-bomb survivor data and notes here, ah, that this raises questions about the general validity of using the worker data for evaluating risks at low doses and dose rates. So, if they don't agree with the A-bomb survivor data, something must be wrong.

Ah, [Seever?] again, talking about paternal effects at Hanford, probably the most important consideration in interpreting these results is the contradictory evidence from other studies, particularly the Japanese A-bomb survivors, this is a third piece of the Japanese studies in addition to the two ah, covered by Dr. Stewart, the Life Span Study of--of long term effects and the study of in utero exposed, this is referring, as Niell's comment here is referring to studies of parents who were exposed and subsequently had offspring. The genetic data of greatest relevance to interpretation of the Gartner Group studies are those collected in the aftermath of the bombings.

So, in summary, we have a situation where the worker studies, despite the fact that they have individual measurement--we're dealing with defined populations who are experiencing exposures that are much more like the ones we're interested in for worker and public protection than the exposures experienced with A-bomb survivors, are essentially being interpreted as having a problem if the results don't agree with the A-bomb survivor data. And I would say, ah, from my perspective, that ah, the ignorance about the work of Stewart and [Niel?] in terms of evaluating evidence in A-bomb survivors for selection effe--effects is--is a primary determinant of this confusion. Um, it is true that the A-bomb studies were initiated before the worker studies. Results there were available earlier. They are dominated by the high dose findings. And they have some statistical properties which are ah--are better in some ways, wh--what scientists often like to see in terms of evaluating dose response, because they have a greater range of doses than the worker studies, but ah, we have now ah, a body of evidence ac--accumulating from worker populations which I believe is now--it's now time to recognize that these studies should be contributing more to the radiation protection standards because of the reasons I enumerated earlier and that in setting these standards, we're responsible to err in the direction of protecting the public and workers and that the use of the worker data w--is very important in this regard. Thank you.